Nous Research Brings Hermes Agent to Microsoft Teams and Azure By Matthew Brienen Published: 2026-11-20T15:45:17+00:00 Nous Research says Hermes Agent is now available in Microsoft Teams, the Microsoft Store and Azure Container Apps Sandboxes. The November 18 announcement puts the agent in widely used workplace tools and adds an Azure option for isolated workloads. Nous Research is an M12 portfolio company, linking the release to Microsoft's venture fund as well as its product ecosystem. Hermes inside the workplace In Microsoft Teams, Nous says employees can assign Hermes knowledge work such as financial analysis, document drafting and inbox triage, then review the finished result in the same conversation. The point is less about opening another standalone AI tab and more about placing the agent in an existing collaboration channel. The release also includes Hermes Desktop for Windows through the Microsoft Store. For workloads that need execution infrastructure, Nous points to Azure Container Apps Sandboxes, which it says run agent tasks in isolated microVMs that start in under a second and release compute when idle. Those details matter for an enterprise buyer because the useful question is not only whether an agent can draft or analyze. It is where it runs, what it can access, how it is separated from other workloads and how people review the output before it becomes part of normal work. The Microsoft Teams integration also changes the review loop. A manager or teammate can see the request and the delivered work in the same place where the task was assigned. That does not remove the need for review, but it may make responsibility clearer than a workflow that jumps between a chat bot, an email thread and a separate execution environment. A shared-skill model Nous is presenting the product around what it calls Collective Wisdom. The company says that when an employee finds a successful way to handle a recurring task, Hermes can turn that approach into a reusable skill made up of a written definition and code for the stable steps. Teammates can then review, propose and share the skill. That is a different pitch from an AI assistant that starts fresh with every prompt. The value depends on whether a company can capture the judgment behind a process, preserve the parts that are repeatable and keep a review step where context still matters. In other words, the product has to become more useful without turning undocumented shortcuts into company-wide automation. A skill library could also become its own governance question. Teams will need to decide who can publish a reusable workflow, what data the workflow can access, when a human review is mandatory and how an outdated procedure is retired. The announcement describes sharing with review, but the real control will be in the organization's operating rules. Nous says that shared intelligence remains a private asset of the organization rather than training data for an AI vendor. That is an important claim for teams handling sensitive work, though buyers will still need to examine the actual tenancy, access controls, retention terms and deployment configuration that apply to their use case. Azure deployment and the enterprise question For organizations with data-residency, security or procurement requirements, Nous says Hermes Enterprise can run in the customer's own Azure tenant. That could be the deciding detail for companies that want agent capabilities but cannot send operational data through a separate hosted environment. The announcement is a product release, not proof of how every enterprise deployment will perform. Still, it shows where the agent market is moving: away from isolated chat tools and toward systems that sit inside collaboration software, execute work in controlled environments and retain a team's approved ways of doing recurring tasks.